CHEN90032 Process Simulation and Control

How CHEN90032 is assessed

What CHEN90032 covers

Twelve chapters take you from a process balance to a tuned, stable control loop -- the full CHEN90032 arc from process dynamics to multivariable control, mapped to the University of Melbourne teaching weeks.

Taught in the Faculty of Engineering and IT, Process Simulation and Control runs in two halves — process dynamics (modelling, Laplace transforms, first- and second-order response) then process control (PID, stability, frequency response, tuning and multivariable control) — and is assessed on four continuous pieces plus a 60% closed-book final that is a hurdle.

This exam-focused guide mirrors CHEN90032 as it is taught and examined at the University of Melbourne, with worked derivations, the tuning rules the exam supplies, and the sign, factor and unit traps that quietly cost marks.

Worked example · free

Routh stability limit, ultimate gain and Ziegler-Nichols PI tuning

Key terms

Transfer function G(s)

The ratio of output to input in deviation form in the Laplace domain, G(s) = Y-hat(s)/U-hat(s), derived with zero initial conditions; it captures how a process responds but carries no absolute-value information.

Steady-state gain K

How far the output ultimately moves per unit input, K = delta-y_ss/delta-u, in output-over-input units; its sign matters (K < 0 means the output falls when the input rises, e.g. a coolant valve).

Time constant tau

How fast a first-order process responds (units of time); a step response reaches 63.2% of its final change at t = tau, 95% at 3*tau and 99.3% at 5*tau.

Deviation variable

A variable measured from its steady-state value (y-hat = y - y-bar); transfer functions act only on deviations, so you subtract the steady state before transforming.

Damping coefficient zeta

The dimensionless shape of a second-order response: zeta > 1 over-damped, zeta = 1 critically damped, 0 < zeta < 1 under-damped (oscillates), zeta < 0 unstable.

Dead time (time delay) theta

A pure transport lag, G = e^(-theta*s); it adds unbounded phase lag (phi = -omega*theta) with an amplitude ratio of 1 (no attenuation), and is what makes a finite ultimate gain exist.

FOPTD model

First-order-plus-dead-time, G = K*e^(-theta*s)/(tau*s + 1) -- the standard low-order model fitted from a step-reaction curve (e.g. by the 1/3-2/3 method).

Characteristic equation

1 + G_OL = 0, where G_OL is the product of all loop elements; its roots are the closed-loop poles, and the loop is stable only if they all lie in the left half-plane.

Routh array / criterion

A tabular test that decides whether every root of the characteristic polynomial has a negative real part -- and hence the loop is stable -- without solving the polynomial; it cannot handle dead time.

Ultimate gain Kcu and period Pu

The proportional-only gain that just sustains a constant-amplitude oscillation, and the period of that oscillation; both feed the Ziegler-Nichols and Tyreus-Luyben tuning rules.

Gain margin and phase margin

GM = 1/AR_OL(omega_c) and PM = 180 deg + phi_OL(omega_g): how much the loop gain can rise, and how much extra phase lag it can absorb, before instability (well-tuned targets GM ~ 1.7-4.0, PM ~ 30-45 deg).

Relative Gain Array (RGA)

Bristol's steady-state interaction measure for a multivariable plant; pair control loops on the relative gain that is positive and closest to 1, and avoid any negative relative gain.

Cascade control

An inner (secondary) loop on a fast intermediate variable nested inside an outer (primary) loop, so fast disturbances are rejected before they reach the primary variable; the inner loop must be faster than the outer.

Feedforward control

Measures a disturbance and acts before it upsets the output; the ideal feedforward controller is Gffc = -Gd/Gp, it does not affect closed-loop stability, and it is always paired with feedback trim.

CHEN90032 FAQ

Questions students ask in each CHEN90032 chapter

What is the difference between the controlled variable, the manipulated variable and the set point?

The controlled variable c is the quantity you want to hold steady — a level, temperature, pressure or composition that a sensor measures. The set point SP (also written R) is the target value you want c to reach. The manipulated variable m is what the controller actually changes — usually a flow or a duty adjusted through a valve — to push c toward SP. The loop works on the error e = SP − c: when c drifts from SP the controller moves m until the error returns to (or near) zero. Chapter 01 · Feedback Control & P&IDs

Why do we shift to deviation variables before finding a transfer function?

Because a transfer function needs zero initial conditions. Subtracting the steady-state equation from the dynamic ODE cancels every constant term and leaves an equation in y′ = y − y̅ that starts from zero, so the Laplace derivative rule ℒ{dy′/dt} = s Y′(s) applies cleanly. Remember to add the steady state back, T(t) = T̅ + T′(t), when you report a real value. Chapter 02 · Dynamic Process Models

Why bother with Laplace transforms instead of solving the ODE directly?

Because it converts calculus into algebra. Transforming a differential equation turns every derivative into a power of s, so the equation becomes a fraction you rearrange for Y(s) and invert with a table — no integrating factors or characteristic-equation gymnastics. It also produces the transfer function G(s)=Y(s)/U(s) that the rest of CHEN90032 (transfer functions, stability, frequency response, controller tuning) is built on. Chapter 03 · Laplace Transforms

What is the difference between the gain K and the time constant τ?

K answers 'how far': it is the ultimate output change per unit input change (Δy_ss/Δu), carrying the ratio of the two engineering units and a sign. τ answers 'how fast': it has units of time and fixes the response speed. A first-order fit must report both, each with its correct unit — they are not interchangeable. Chapter 04 · Transfer Functions & First-Order Systems

Is second-order dynamics a big part of the CHEN90032 exam?

It sits inside the transfer-function and frequency-response material, which was a 35-mark question in a representative past paper. The final exam is worth 60% and is a hurdle you must pass to pass the subject: 4 questions / 100 marks, 15 minutes reading plus 3 hours writing, closed-book with a provided formula sheet (Laplace-transform table, tuning rules, Padé approximants, Routh arrays and the Smith's-method chart) and a Casio FX82 calculator. Note that the overshoot, decay-ratio, peak-time and period formulae are NOT on the provided sheet, so memorise them. Confirm the exam date on Canvas — this unit next runs in Semester 1, 2027 (S1 exam period, around June). Chapter 05 · Second-Order & Complex Process Dynamics

What is the difference between dead time and inverse response?

Dead time means the output does nothing for a period θ and then moves the right way; inverse response means the output moves immediately but in the wrong direction first before turning around. Dead time comes from a transport delay e^(−θs); inverse response comes from a right-half-plane (positive) zero. Both reduce phase margin and make control harder, but only a right-half-plane pole (not a zero) causes actual instability. Chapter 06 · Time Delay, Integrating Processes, Poles & Zeros

Why does proportional-only control leave an offset while PI does not?

With P-only control the manipulated output is proportional to the error, so it needs a non-zero error to hold the valve away from its bias — the loop settles with a residual offset of 1/(1 + K_cK_p). Integral action keeps changing the output for as long as any error exists, so it only stops when the error is exactly zero. The cost is that the integrator adds a pole at s = 0, raising the closed-loop order by one, which can reduce the stability margin. Chapter 07 · Closed-Loop Behaviour & PID Controllers

If every coefficient of the characteristic polynomial is positive, is the loop stable?

Not necessarily. All coefficients being present and positive is a necessary condition for stability, but from third order upward it is not sufficient — a cubic or quartic with every coefficient positive can still have roots in the right-half plane. For a first- or second-order polynomial the positive-coefficient test alone does guarantee stability, but for anything higher you must build the Routh array (or apply the collapsed 3rd- and 4th-order inequalities) and confirm the whole first column is positive. If a coefficient is missing or negative you can stop immediately: the loop is already unstable. Chapter 08 · Control-System Stability & Routh Analysis

Why is stability decided at the frequency where the phase is −180°?

At the phase-crossover frequency ω_c the feedback signal returns exactly inverted, so it reinforces a fresh disturbance instead of cancelling it. If the loop also has enough gain there (AR_OL(ω_c) ≥ 1) the oscillation sustains or grows and the closed loop is unstable. If AR_OL(ω_c) < 1 the returning signal is smaller each cycle and the loop is stable — that is the whole Bode criterion. Chapter 09 · Frequency Response, Bode Plots & Stability Margins

How do I get the ultimate gain Kcu and ultimate period Pu without running a plant to instability?

Two safe routes. From a model, compute them by frequency response: find the critical frequency ωc where the open-loop phase equals −180°, then Kcu = 1/AR_p(ωc) (the reciprocal of the process amplitude ratio there) and Pu = 2π/ωc. Experimentally, use relay auto-tuning: an on-off element forces a small, bounded limit cycle whose period is Pu, so you never push the plant to true instability. Note that a pure first- or second-order lag with no dead time has no finite Kcu at all — it is the time delay (or three-plus lags) that creates a stability limit. Chapter 10 · PID Controller Tuning

How is direct synthesis different from Ziegler–Nichols or Tyreus–Luyben tuning?

Ziegler–Nichols and Tyreus–Luyben are empirical rules: you find the ultimate gain K_cu and ultimate period P_u (from a Bode plot or a relay test) and read the controller constants off a table. Direct synthesis is model-based: you write down the process transfer function and the closed-loop response you want, then solve algebraically for the controller. For a first-order-plus-time-delay process it produces a PI/PID with K_c = τ/[K(τ_c + θ)] and T_I = τ, and its one knob τ_c lets you dial the closed-loop speed directly — at the cost of needing an accurate model. All three appear on the tuning side of the subject; the provided formula sheet lists the Ziegler–Nichols and Tyreus–Luyben rules and the Smith predictor. Chapter 11 · Advanced Control: Direct Synthesis, Delay Compensation, Cascade, Feedforward, Ratio

How do I know which loop to pair with which manipulated variable?

Compute the Relative Gain Array from the steady-state gains and pair each controlled variable with the manipulated variable whose relative gain λ is positive and closest to 1. A value between 0.5 and 1 is a good pairing with mild interaction; a value below 0.5 means the indirect path dominates; and a negative λ is dangerous because the effective gain changes sign when the other loops are switched to automatic, so it must never be paired. Chapter 12 · Multivariable (MIMO) Control: RGA & Decoupling

How to study for the exam

Treat CHEN90032 as one chain and drill it end to end: a balance becomes an ODE, an ODE becomes a transfer function, and a transfer function becomes a tuned, stable loop. Spend the first weeks making the dynamics automatic -- deviation variables, the Laplace table, first-order K and tau, the second-order descriptors (overshoot, decay ratio, period) and the FOPTD fit -- because every control question is built on them.

Then master the control toolkit as a sequence: the closed-loop transfer functions and characteristic equation, the Routh and Bode stability tests, the ultimate gain and period, and the Ziegler-Nichols and Tyreus-Luyben tuning rules, finishing with the advanced and multivariable structures (cascade, feedforward, direct synthesis, RGA and decoupling).

Practise under the real constraints -- closed-book, a Casio FX82, and 1.8 minutes per mark -- always writing the method and carrying the sign and units, since the final is a hurdle you must pass and it awards working, not just the final number.

Use SWOTVAC -- the study week between the end of teaching and the exam period -- for final revision: work full past exam papers and re-authored practice to time, closed-book with the provided formula sheet, so the method is automatic on the day. When a step will not click, ask Sia to explain it, then redo the problem from a blank page.

Your AI Engineering tutor for CHEN90032

Stuck on a hard CHEN90032 question? Sia is AskSia’s AI Engineering tutor — ask any CHEN90032 Process Simulation and Control question and get a clear, step-by-step explanation grounded in how the course is actually taught and assessed. Read this whole study guide free, then take your hardest questions to Sia.